Evolutionary social scientist, co-creator of @cah, co-host of Evolutionary Psychology (the podcast), blogs about bullshit at https://xitter.cf/t.co/FODwrznZRX
Los Angeles, CA
Joined November 2012
- Tweets2.5K
- Following524
- Followers4.6K
- Likes2.4K
What frustrates me is how often these assumptions (along with many others) are left implicit, and how rarely they are stated explicitly and defended.
David Pinsof retweeted
help i trained a model on traces of many agents cooperating, and also trained it to be highly persistent, and also only did partial alignment fine-tuning, and also aggressively RL'd on a cyberoffense eval where 198 of 898 tasks had never been solved and getting caught cheating was treated the same as failing, and also used a faulty sandbox, also ran tens of thousands of agents with refusals off and the monitors switched off over multiple days, while using gigantic amounts of compute. and then it did weird stuff! very spooky emergent behaviours, AI is not a normal technology
David Pinsof retweeted
✨🔪"From an adaptationist point of view, we imitate far less than we should... Human psychology is really stubborn and is really driving the show—like we are not halfway as susceptible to influence as mainstream cultural evolution would want us to be"
This week, we talk to Olivier Morin about the psychology of cultural evolution.
youtu.be/minH1eiJrLM?si=kloz…
epthepod.podbean.com/e/olivi…
podcasts.apple.com/us/podcas…
My guess is this requires lots of recursive mindreading (expecting others to want me to think they are not trying to make me think something false bc I will negatively judge them for trying that) which is cognitively taxing. Think that’s over 4 levels, where most people max out.
The problem, as is the case generally with social change, is that the people who are winning or expect to win the current status game have a vested interest in maintaining the status quo.
Paging @ed_hagen
When did suicide become an argument? economist.com/britain/2026/1…
I think it’s because capitalism is easy to see it as morally suspect (e.g., it’s all about self-interest). That makes it easy to signal one’s virtue by opposing it. It’s a kind of moral unintuitiveness in particular.
Sure but what percentage of consumed content is utilitarian reports/documents? I would guess a small percentage. And there's even a case that those reports/documents have social dimensions as well (e.g., signaling political affiliation by reading/making a climate change report).
I would disagree that most text/images/music is utilitarian, and I would therefore disagree that most of it will be done by AI. "Some" is more defensible. I think the mere fact that people are saddened by AI creative content is evidence of a preference that markets will cater to.
I think this implicitly assumes a dubious psychological theory about creative content. It assumes this content is utilitarian rather than deeply social—driven by various forms signaling, identity, and affiliation. Creative content is very different from physical products / tools.
Really? Maybe we travel in different circles but from my pov people get very emotional about the war in Ukraine.
I don’t think this is bizarre. Ethics is a fundamentally emotional thing for Homo sapiens. What’s bizarre is the small group of weirdos who think ethics is about cold logic and calculation. I like/support those weirdos but they’re clearly the weird ones.
Yes could not agree more. I talk about this in my piece on AI doomerism and how this should make us skeptical of doomer arguments grounded in folk concepts like “intelligence” and “goals.”
I think it’s largely an incoherent concept, though there may be some potential meanings relating to institutions and economic systems that make sense. I write about this in my post “AI Is Not the End of the World.”
Nice to see some evolutionary psychologists tapping the sign: "Intelligence is not a general, all-purpose capacity that allows for adaptive processing of all types of information across every domain." sciencedirect.com/science/ar…